Generalization of the Strong Interest Inventory in Chinese ...

St. Cloud State UniversitytheRepository at St. Cloud State

Culminating Projects in Psychology Department of Psychology

5-2017

Generalization of the Strong Interest Inventory inChinese Culture A Study of Translation, Validation,and Cross-Cultural ComparisonsYang YangSt. Cloud State University, [email protected]

Follow this and additional works at: https://repository.stcloudstate.edu/psyc_etds

This Thesis is brought to you for free and open access by the Department of Psychology at theRepository at St. Cloud State. It has been accepted forinclusion in Culminating Projects in Psychology by an authorized administrator of theRepository at St. Cloud State. For more information, pleasecontact [email protected].

Recommended CitationYang, Yang, "Generalization of the Strong Interest Inventory in Chinese Culture A Study of Translation, Validation, and Cross-CulturalComparisons" (2017). Culminating Projects in Psychology. 7.https://repository.stcloudstate.edu/psyc_etds/7

https://repository.stcloudstate.edu?utm_source=repository.stcloudstate.edu%2Fpsyc_etds%2F7&utm_medium=PDF&utm_campaign=PDFCoverPages

https://repository.stcloudstate.edu/psyc_etds?utm_source=repository.stcloudstate.edu%2Fpsyc_etds%2F7&utm_medium=PDF&utm_campaign=PDFCoverPages

https://repository.stcloudstate.edu/psy?utm_source=repository.stcloudstate.edu%2Fpsyc_etds%2F7&utm_medium=PDF&utm_campaign=PDFCoverPages

https://repository.stcloudstate.edu/psyc_etds?utm_source=repository.stcloudstate.edu%2Fpsyc_etds%2F7&utm_medium=PDF&utm_campaign=PDFCoverPages

https://repository.stcloudstate.edu/psyc_etds/7?utm_source=repository.stcloudstate.edu%2Fpsyc_etds%2F7&utm_medium=PDF&utm_campaign=PDFCoverPages

mailto:[email protected]

Generalization of the Strong Interest Inventory® in Chinese Culture

A Study of Translation, Validation, and Cross-Cultural Comparisons

by

Yang Yang

A Thesis

Submitted to the Graduate Faculty of

St. Cloud State University

in Partial Fulfillment of the Requirements

for the Degree of

Master of Science

in Industrial and Organizational Psychology

May, 2017

Thesis Committee:

Dr. Daren Protolipac, Chairperson

Dr. John Kulas

Dr. Richard Thompson

2

Abstract

The current 2004 revision of the Strong Interest Inventory has been understudied in China. The

present study (a) translated the Strong assessment into Simplified Chinese, (b) investigated the fit

of circular and circumplex models of Holland’s theory in Chinese population and compared the

scores on the construct equivalent scales, and (c) uncovered the generalizability and applicability

of the Strong assessment in Chinese culture. The randomization test (RTHOR) and circumplex

covariance structure model (CCSM) were applied to a diverse Chinese sample to explore the

cross-cultural validity of Holland’s models. Empirical support was found for Holland’s circular

ordering model in the overall sample and subgroups of males and students. Results suggested

that the Chinese Strong assessment was psychometrically sound and was promising to be used in

China. Theoretical and practical implications were then discussed.

Keywords: Strong Interest Inventory; Generalization; Translation; China

3

Acknowledgement

I would like to thank Mike Morris for his kindest provision of enormous help and valuable

feedback, and Daren Protolipac, John Kulas, and Richard Thompson for their constructive

comments on this study. I would also like to extend my sincerest thanks and appreciation to Yun

Wu, Ming Zhu, Zhonghong Yang, Xinliang Li, Ying Liu, Beilei Pan, Fanxin Zhu, and other

individuals for helping recruit and organize participants in China to take the Chinese Strong

Interest Inventory. In addition, I owed much to Yuzhou Chen, Qiuyu Su, and Linda Fan, who made

great effort to translate and review the Strong Interest Inventory. Lastly, I have to specially thank

the CPP, Inc., the exclusive publisher of the Strong Interest Inventory, for providing me with great

trust and opportunity to translate and validate the Strong assessment.

4

Table of Contents

Abstract ......................................................................................................................2

Acknowledgement .....................................................................................................3

Chapter I: Introduction ...............................................................................................8

Chapter II: Literature Review ....................................................................................12

Overview of Holland’s Theory of Vocational Personalities ..........................12

Research of Holland’s Model in China ..........................................................14

The 1994 Chinese Revision of the Strong Interest Inventory ........................15

Chapter III: Method ...................................................................................................19

Participants .....................................................................................................19

Instrument ......................................................................................................19

Administration ...............................................................................................20

Data Cleaning.................................................................................................21

Analyses .........................................................................................................22

Chapter IV: Translation and Adaptation Procedures .................................................25

The Translation and Review Committee .......................................................25

First Phase: Direct Translation.......................................................................26

Second Phase: Review and Reconciliation ....................................................26

Chapter V: Results .....................................................................................................28

Descriptive Statistics and Reliability .............................................................28

Randomization Test .......................................................................................28

Circumplex Covariance Structure Modeling .................................................29

Comparisons between Two Cultures .............................................................30

5

Chapter VI: Discussion ..............................................................................................31

Theoretical implications.................................................................................32

Practical Implications.....................................................................................33

Limitations and Future Research ...................................................................33

Chapter VII: Conclusion ............................................................................................35

References ..................................................................................................................36

Appendix A: Tables ...................................................................................................47

Appendix B: Figures ..................................................................................................62

Appendix C: IRB Review ..........................................................................................69

6

List of Tables

Table 1. Summary of Research on Testing Holland’s Model in China .....................47

Table 2. Suggestions from Studies on Testing Holland’s Model in China ................48

Table 3. Demographic Information of Participants ...................................................49

Table 4. GOT Reliability Statistics ............................................................................50

Table 5. Correlation Matrix for the Overall Sample (N = 364) .................................51

Table 6. Correlation Matrix for the Male Group (N = 135) .......................................52

Table 7. Correlation Matrix for the Female Group (N = 229) ...................................53

Table 8. Correlation Matrix for the Student Group (N = 191) ...................................54

Table 9. Correlation Matrix for the Employee Group (N = 127) ...............................55

Table 10. BIS Reliability Statistics ............................................................................56

Table 11. PSS Reliability Statistics ...........................................................................58

Table 12. Results of Randomization Test of Hypothesized Ordering Relations .......59

Table 13. Model Fit Statistics for Circumplex Covariance Structure Modeling .......60

Table 14. Point Estimates and Confidence Intervals for Polar Angles and

Communalities .......................................................................................................................61

7

List of Figures

Figure 1. Categorization of four specific Holland’s models ......................................62

Figure 2. Visual presentation of four specific Holland’s models ..............................63

Figure 3. Unconstrained (circular ordering) model for the overall sample ...............64

Figure 4. Unconstrained (circular ordering) model for the male group .....................65

Figure 5. Unconstrained (circular ordering) model for the female group ..................66

Figure 6. Unconstrained (circular ordering) model for the student group .................67

Figure 7. Unconstrained (circular ordering) model for the employee group .............68

8

CHAPTER I: INTRODUCTION

Research on the cross-cultural validity of Holland’s theory of vocational personality

types has been prevalent in the field of vocational psychology for several decades (Bullock,

Andrews, Braud, & Reardon, 2009; Day & Rounds, 1998; Farth, Leong, & Law, 1998; Fouad,

1993; Leong, Austin, Sekaran, & Komarraju, 1998; Leong, Hartung, & Pearce, 2014; Rounds &

Tracey, 1996; Subich, 2005). As the majority of popular Holland-based inventories in use are

were developed on the population makeup of U.S. society (e.g., Self-Directed Search, Strong

Interest Inventory), a frequently asked question for the international use of these inventories is

whether Holland’s RIASEC model retains the same structure and ordering in non-Western and/or

non-English speaking countries. Many have underscored the importance of carefully examining

construct equivalence of the model before directly comparing scale scores to culturally different

individuals (Long & Tracey, 2006) and interpreting the profiles without any consideration of

cultural factors (Fouad, 1993; Westermeyer, 1987).

Language is, if not the only, the fundamental disparity when conducting cross-cultural

research (Geisiger & McCormick, 2013). To overcome the language barrier, psychologists and

linguists have made joint efforts to translate and adapt the vocational interest assessments into

local languages and to examine the generalizability of Holland’s theory in countries outside of

the U.S. (e.g., Glidden-Tracey & Greenwood, 1997; Goh & Yu, 2001; Hansen & Fouad, 1984).

As for China, the increasing attention has also been paid on the transportability of Holland’s

theory in Chinese society to meet scientific and societal inquiry (Fan & Leong, 2016). Several

trending Western-based interest inventories have been translated into Simplified and/or

Traditional Chinese and validated by using local Chinese populations (Goh & Yu, 2001; Wang,

Xue, Li, & Zhang, 2016; Yang, Lance, & Hui, 2006; Zhang, Wei, Li, & Wang, 2015). These

9

endeavors, however, suggested mixed support for the fit of Holland’s models in Chinese culture,

which may be further explained by many factors such as incomparable quality of inventory

translation, different choices of inventory in use, varied sample constitutions, and so forth.

Therefore, vocational psychologists and practitioners have been calling for more investigations

into cross-cultural validity (or construct equivalence) of Holland’s theory in Chinese context

(Fan & Leong, 2016; Hao, Sun, & Yuen, 2015; Yan, 2008).

The Strong Interest Inventory (Donnay, Morris, Schaubhut, & Thompson, 2004) is one of

the most popular vocational assessments used in the U.S. and many other countries. Much

evidence has been found in the literature that the Strong assessment is reliable and valid to use

regardless of race, ethnicity, and/or country of origin (Armstrong, Hubert, & Rounds, 2003;

Fouad, Harmon, & Borgen, 1997; Kantamneni & Fouad, 2011, Kantamneni, 2015). A most

recent meta-analysis study (Nye, Su, Rounds, & Drasgow, 2017) also concluded that the Strong

Interest Inventory outperformed other popular vocational assessments (i.e., Self-Directed Search,

Vocational Preference Inventory, and Kuder Preference Recode) when it was used to predict

performance criteria such as task performance and organizational citizenship behaviors.

Interestingly enough, a literature search of five major journals in the field (i.e., Journal of

Counseling Psychology, Journal of Vocational Behavior, Journal of Career Assessment, Career

Development Quarterly, and Journal of Career Development) and the most authoritative Chinese

journal database (i.e., CNKI) revealed only five articles focusing on the Strong assessment in the

Chinese context. All of them used previous revisions of the Strong assessment – three in English

(Goh, Lee, & Yu, 2004; Goh & Yu, 2001; Tang, 2001) and two in Chinese (Chen & Shen, 1997;

Ge, Yu, & Wang, 1996). As the most ubiquitous vocational interest assessment, the current

revision of the Strong assessment (Donnay et al., 2004) has, ironically, never been explored in

10

China or with respect to Chinese culture, which becomes the important impetus of the current

study.

In line with the consideration of cultural validity of career assessments (Leong & Brown,

1995; Marsella & Leong, 1995) and in response to aforementioned research needs, the objective

of this study is three-fold: (a) to translate and accommodate the latest Strong Interest Inventory

into Simplified Chinese, (b) to investigate the cross-cultural validity of Holland’s RIASEC

models in Chinese culture, and (c) to evaluate the transportability and generalizability of the

Strong assessment in the Chinese population. This study has research and practical implications

for the expansion of Holland’s theory in a typical non-Western and non-English speaking

country. Furthermore, it may also pave the way for future research examinations and benefit

practitioners (e.g., career counselors) from expanding the availability of vocational assessment

tools in China. To the author’s knowledge, this study is the first to result in a Simplified Chinese

form of the 1994 Strong assessment.

The structure of the remaining sections is arranged as follows. The literature review

section overviews Holland’s theory and four specific models, as well as previous research on

Holland’s RIASEC models in Chinese populations by various interest inventories. Particular

emphasis is placed on the translation and field-testing studies of the Strong Interest Inventory

(1994 Chinese revision). This is followed by the method section describing the sample

composition, instrument, administration, data cleaning and analysis procedure. In contrast to

traditional research articles, the translation procedure details are clarified in a subsequent section

title “Translation and Adaptation Procedures”, which includes an overview of the “forward

translation” approach, the translation and review committee composition, and item translation

specifications. The results section presents numbers, tables, figures, and narratives that

11

demonstrate the fit of Holland’s models and reliability and validity evidence for the Chinese

Strong assessment. This study culminates with research and practical implications, as well as

limitations and recommendations.

12

CHAPTER II: LITERATURE REVIEW

Overview of Holland’s Theory of Vocational Personalities

Holland’s (1973, 1985, 1997) categorization of people’s interest (or personality) into six

types – Realistic, Investigative, Artistic, Social, Enterprising, and Conventional – is probably the

most influential typological framework in vocational psychology (Lowman & Carson, 2013).

The overarching assumption is that correlations between two adjacent interest types are greater

than those between alternate types and in turn greater than those between opposite types. Two

hypotheses derived from Holland’s work (1973, 1985, 1997) have resulted in different models

examined by subsequent researchers. The calculus hypothesis posits that six interest types are

manifested in a circular order and the distance between either two types are “inversely

proportional to the theoretical relationships between them” (Holland, 1973, 1985, 1997, p. 5),

which was later evolved the circular ordering model (or circular order hypothesis, Rounds,

Tracey & Hubert, 1992). While the hexagonal hypothsis derived from Holland (1973, 1985,

1997) specifies that six interest types are shaped into an equilateral hexagon where distances

between adjcent types are equal. This unique arrangement of types resembles Guttman’s (1954)

circumplex model of personality (for a detailed discussion, see Hogan, 1983).

More recent literature (e.g., Darcy & Tracey, 2007) in the field suggests that there are

four specific models based on Holland’s work, which are determined by two parameters (Figure

1): angular locations (i.e., the polar angles between two types) and communalities (vector length

of each type) in the circle (Morgan & Bruin, 2017). Figure 2 provides an explicitly visual

presentation of these four models and their differences. The circular ordering model (the least

restrictive) and circumplex model (the most restrictive) are derived from Holland’s calculus and

hexagonal hypothese, which are most popular ones examined by many vocational psychologists

13

(Rounds et al., 1992; Sodano, 2015). The former focuses on the RIASEC order of types (Tracey,

2000), while the latter examines the equidistance of types on a circumference (Darcy & Tracey,

2007). Falling between them are two quasi-circumplex models constrained by one parameter. In

other words, one quasi model probes whether interest types have the same vector in the circle

and set the angular locations free, while the other superficially investigate whether polar angles

between adjacent interest types equal 60 degrees.

Although Holland’s model is proposed based on U.S. populations (Holland, 1973, 1985,

1997), much attention has also been put on testing model’s applicability in other countries and

cultures. Previous research indicated that the circular ordering model received more support from

U.S groups (Kantamneni & Fouad, 2011; Kantamneni, 2014) than from various international

samples (Rounds & Tracey, 1996). While contradictory evidence was found for quasi-

circumplex and circumplex models across U.S. ethnic groups (Day & Rounds, 1998; Day,

Rounds, & Swaney, 1998; Tracey & Robbins, 2005), as well as culturally diverse groups (e.g.,

Morgan & de Bruin, 2017, in Africa), in that these models are more stringent than the circular

ordering model.

Another line of research has examined the potential differences across sex and how such

differences influence the ordering of RIASEC model and scores on each interest type. Although

previous research provided evidence that males and females shared the same RIASEC order

(e.g., Darcy & Tracey, 2007, Tracey & Robbins, 2005; Tracey & Rounds, 1993), other studies

did not corroborate these findings (Anderson, Tracey, & Rounds, 1997; Kantamneni & Fouad,

2011; Kantamnenei, 2014; Morris, 2016).

Su, Rounds, and Armstrong’s (2009) meta-analysis found that men scored higher on

Realistic and Investigative interests, whereas women had higher scores on Artistic, Social, and

14

Conventional interests. A more recent primary study (Morris, 2016) examining 1,283,110 U.S.

residents who completed the Strong assessment revealed that substantial sex differences across

age and ethnic groups and such differences were consistent over the period from 2005 to 2014,

with the exception that people between 18 and 22 years old showed slightly sex differences in

more recent samples.

Research of Holland’s Model in China

Examinations of Holland’s theory and model in China or Chinese culture are not found to

be dominant in vocational literature. Only one meta-analysis published in the last decade (Long

& Tracey, 2006) summarized the structure of RIASEC scores in China by evaluating the fit of

four particular models: Holland’s circular order model1; Gati’s three-group partition model;

Rounds and Tracey’s alternative three-group partition model; and Liu and Rounds’ modified

octant model on 29 correlation matrices collected from 13 empirical studies. It was concluded

that Holland’s model had the worst fit in the Chinese population among four models and had a

lower fit than in the U.S. samples. However, this synthetic finding may be skeptical to be applied

to the contemporary Chinese society because sources of correlation matrices were derived from

research between 1987 and 2001 and the majority of the samples were student groups at all level

(middle school, high school, and college). Therefore, the author searched for and examined more

recent literature (esp. 2000 and later) and enumerated the major findings in Table 1.

Generally speaking, studies in Table 1 paint a contradictory picture in terms of the

applicability of Holland’s model in several native Chinese samples. The Strong Interest

Inventory (SII) and Self-Directed Search (SDS) were popular instruments that were used to

conduct the cross-cultural validation studies. Almost all research have provided mixed evidence

1 This is what is called “Circular Ordering Model” in this study.

15

(Goh & Yu, 2001; Tang, 2001; Yang, Stokes, & Hui, 2005; Tang, 2009) or no support (Goh et

al., 2004) for Holland’s circular or circumplex model across geographic groups in China through

various analytical approaches, such as correlation, exploratory factor analysis (EFA), and

confirmatory factor analysis (CFA). Furthermore, sex differences were evident to the structure

and/or ordering of six interest types in most of the existing research (Goh & Yu, 2001; Tang,

2001; Yang et al., 2005; Tang, 2009). Researchers of these studies have called for further

investigation and replication of interest inventories to diverse, large-scale, and representative

Chinese samples (Table 2) before any general conclusions can be drawn. The following section

discusses research conducted on the Chinese revision of SII in detail as it is the focus of this

study.

The 1994 Chinese Revision of the Strong Interest Inventory

Among all previous revisions of the Strong Interest Inventory, the 1994 revision

(Harmon, Hansen, Borgen, & Hammer, 1994) is the only one that was translated into Simplified

Chinese. Therefore, a brief overview of the translation procedure (Ge et al., 1996) and

subsequent research (Goh et al., 2004; Goh & Yu, 2001; Tang, 2001) is given to this revision.

Ge, Yu, and Wang (1996) made the first attempt to translate the Strong Interest Inventory

(1994 revision) into Simplified Chinese as a portion of a cross-cultural research project between

China and the U.S. The translation panel was comprised of six (three in China and three in the

U.S.) professionals who were proficient in language and culture of both countries as well as

basic knowledge of psychological testing and assessments. A rigorous three-step procedure (i.e.,

direct translation, back translation, and reconciliation) was strictly followed and resulted in 302

items (95.3% of 317) achieving linguistic and inferential equivalence. The remaining 15 items

with no linguistic equivalence were replaced by comparable translation items.

16

This translation revision was subsequently tested and validated by three studies surveying

different Chinese samples. Goh and Yu (2001) conducted a field test based on two Chinese

samples (N1 = 124, N2 = 40) and one American sample (N3 = 52). The metric equivalence of the

translation was found through correlations, t-tests, and profile analyses between two Chinese and

one American samples. Results of EFA suggested that three of six factors approximated the

interest types of Artistic, Realistic, and Social, while the other three were deviant from the

original classifications. Furthermore, they relabeled one factor as “Public” rather than the

original “Conventional”, in that Basic Interest Scales in this factor is more relevant to public

affairs. The same year, Tang (2001) administered the Chinese Strong assessment to 166 college

students enrolled in several Chinese universities and explored Holland’s model through MDS.

Results suggested that males and females had similar but not identical RIASEC orders (RISAEC

for males and RSAECI for females). They then conducted an EFA on 25 Basic Interest Scales

and found that factors that were extracted did not resemble the original classification. Later, Goh,

Lee, and Yu (2004) surveyed 247 Chinese high school students using the Chinese Strong

assessment. The CFA findings revealed that the sample did not fit Holland’s six-factor model. In

addition, their direct examination of intercorrelations among the factors provided weak evidence

for the circular ordering of six interest types hypothesized by Holland. Alternatively, the EFA

suggested a three-factor solution2 that mirrored the underlying structure of the Chinese Strong

assessment with sufficient amount of variance accounted for (Goh et al., 2004).

Obviously, none of these validation studies provide sufficient evidence that the structure

and ordering of Holland’s model are applicable to the Chinese population and the Chinese SII

can be potentially used in China. Three major limitations concerning translations, samples, and

2 Factor 1: Artistic/Social; Factor 2: Enterprising/Conventional; Factor 3: Realistic/Investigative

17

methodologies may contribute to the equivocal findings across the research. The first limitation

is that there are 15 items that lack linguistic and inferential equivalence, which are culturally

relevant for Chinese people. These items in Chinese culture may not carry the identical

conceptual message as they are expected, even replaced by comparable translations. Therefore,

discrepancies in item meanings probably decrease applicability of Holland’s model in the

Chinese population. The second limitation appearing in all three studies concerns the samples

that are characterized by small sizes and a lack of diverse. Specifically, the sample sizes are

below the recommended minimum subject-to-item ratio of 5:1 (Bryant & Yarnold, 1995;

Gorusch, 1983, p.332) or even N ≥ 300 for factor analysis (Comfrey & Lee, 1992, p.127). As

described above, samples were comprised of high school or college students, which make it hard

extrapolate the conclusions to non-student groups such as working adults.

The third limitation is about inappropriate analytical approaches and procedures that were

applied to validate the underlying structure and ordering of Holland’s model. One the one hand,

for example, MDS (Tang, 2001) and correlation analysis (Goh et al., 2004) were implemented to

test the calculus hypothesis of Holland’s model. Specifically, the MDS involves obviously

subjective judgment that extracts two dimensions to visualize the circumplex structure3 without

statistics that are crucial to indicate the goodness of fit (Fabinger, Visser, & Browne, 1997). In

addition, direct observations of correlation matrices without visual aid (Goh et al., 2004)

produced more judgmental errors regarding the calculus hypothesis. On the other hand,

inappropriate analytical procedures that compare the scale scores without warranted structure

equivalence (e.g., Goh & Yu, 2001) may generate questionable and misleading conclusions from

a methodological perspective.

3 The circumplex structure considers (a) ordering, (b) angular locations, and (c) communalities among interest types.

18

To summarize, although the applicability of Chinses SII was not empirically supported

by the existing literature, a great deal of effort on translation and validation of the 1994 Strong

assessment provides insights into translation, sampling, and methodology for the current study

that continue to explore the cross-cultural validity of the most recent Strong assessment.

19

CHAPTER III: METHOD

Participants

The current sample contained 364 native Chinese participants whose country of origin

and residence were People’s Republic of China. They were asked to fully complete the Chinese

Strong assessment for personal development purpose. Table 3 shows the demographic

information of these participants. There were about twice as many females (N = 229) compared

to males (N = 135) in the sample. The average age of all participants was 24.09 years (SD =

6.70, median = 23.00), ranging from 15 to 50 years. Two hundred and forty-five participants

(67.30%) hold a bachelor’s degree or higher. One hundred and ninety-one participants self-

identified as full-time students (age M = 20.18, SD = 2.74) and 127 as full-time employees (age

M = 29.83, SD = 7.02). Among full-time working adults, 102 were entry-level or non-

supervisory employees and 31 were at supervisor level or higher.

Instrument

The Strong Interest Inventory is a highly regarded career assessment tool most commonly

used for helping individuals make educational and occupational choices (Donnay et al., 2004).

The current revision of the Strong Interest Inventory has 291 items that assess interest in

occupations, specific areas of school subjects, activities, people, and personal characteristics on a

5-point Likert-type option anchored by Strong Like to Strongly Dislike. Responses are

standardized into four board categorizations with a mean of 50 and standard deviation of 10

based on a General Representative Sample (GRS, or normative group) consisting of 2,250

respondents (50% male, 50% female, diverse with regard to age and ethnicity) that represent the

adult U.S. workforce. The General Occupation Themes (GOTs) are the operationalization of

Holland’s interest types, with 21 to 31 items for each theme (α = .90 to .95, median = .92). Thirty

20

Basic Interest Scales (BISs) provides more specific domains that are composed of homogeneous

items, with 6 to 12 items for each scale (α = .80 to .92, median = .87). Occupational Scales are

the most specific ones that reflect similarities between respondents and people who are employed

in and satisfied with particular occupations4. Personal Style Scales (PSSs) demonstrate people’s

living and working styles, with 9 to 41 items for each scale (α = .82 to .87, median = .86).

Subsequent research with large samples has provided adequate evidence for the concurrent

validity and counseling utility (Gasserm Larson, & Borgen, 2007) and the structure equivalence

across races and ethnicities in the U.S. (Kantamneni & Fouad, 2011; Kantamneni, 2014).

Another technical brief (Herk & Thompson, 2011) concluded that the Strong assessment has

similar and comparable results across translation versions of European English, French, German,

Latin American Spanish, and European Spanish.

Administration

Instructions and items of the Chinese Strong assessment were loaded onto a leading

online survey platform by the author who had access to a secured account. The snowball

sampling technique (Goodman, 1961) was used to collect the email address of potential

participants through author’s personal network back in China. A total of 966 email invitations

were sent to people who showed interest in the assessment and 441 participants (45.7% response

rate) completed it. An electronic informed consent was presented before participants moved

forward to respond the interest items. Participants were also informed that only those who

finished all 291 items in the inventory could get a well-developed standardized feedback report

generated by the author via email, albeit response to all items was not required to create reports

4 Because Cronbach’s alphas are not given in the Strong technical manual (Donnay et al., 2004) and OSs are not the

focus of this study, α values are not reported here.

21

(Donnay et al., 2004, p.159). This notice acted as an incentive for participants to go through each

question with their whole attention as well as a mechanism that naturally selected out

participants who were not willing to complete the inventory. Due to the length of the assessment,

participants were allowed multiple accesses to complete all items within one month. Several

actions were further taken to protect the copyright of the assessment and secure the data

collected from participants.

Data Cleaning

Despite the incentives to complete the inventory, careless responses cannot be avoided

given 291 items to respond. Therefore, the data cleaning procedure was applied to 441

participants to identify potentially bad cases characterized by inconsistent item endorsements and

irregular response patterns. In particular, the typicality index (Donnay et al., 2004, p.159),

designed to catch people who respond in a random fashion, was utilized to help indicate

participants who endorsed items in an unusual manner by summarizing the combination of

responses of 24 item pairs. The typicality index ranges from 0 (no consistent responses) to 24 (all

pairs responded to consistently), and a score lower than 17 indicated possibly inconsistent

responses. One respondent had a typicality index lower than 17 and was excluded. Irregular

response patterns can also be recognized via looking at the response percentages of five response

options (e.g., indifferent). Although normal ranges of possible response percentages for GRS was

provided in the Strong technical manual (Donnay et al., 2014, pp.153-158), these criteria cannot

be directly applied to culturally different individuals because they may have different response

styles in answering items (Van de Vijver, 2000, Van de Vijver, 2015). Therefore, an exploratory

cutoff score of 70% was applied to the sample in this study. That is, if one participant endorsed

the same option across over 203 items (70% of 291), he or she was considered to complete the

22

assessment without paying enough attention. Seventy-six participants violated the 70% cutoff

and were excluded, leaving a total of 364 respondents.

Analyses

The correct analysis procedure in cross-cultural studies is specifically important because

between-group mean comparisons based on non-equivalent scales and measures are skeptical

and problematic and are more inclined to result in misleading and meaningless interpretations

and conclusions (Long & Tracey, 2006; Rounds & Tracey, 1996; Fouad, 2002). This implication

is usually neglected by cross-cultural researchers. In the light of this, the current study adhered to

a restrictive analysis procedure that (a) reliability (Cronbach’s alpha) was first calculated to

examine the internal consistency of responses on GOT, BIS, and PSS scales, followed by (b)

randomization test of hypothesized order relations as well as circumplex covariance structure

modeling that evaluated four specific RIASEC models and visually present them in circles, and

finally, (c) scale scores of GOT, BIS, and PSS were compared against the U.S. normative scores.

Note that only when metric equivalence is achieved can the scores be compared across groups.

The following part briefly introduces the randomization test and circumplex covariance structure

modeling.

The randomization test of hypothesized order relations (RTHOR; Hubert & Arabie, 1987)

has emerged to become a better method (Rounds et al., 1992) and is frequently used to evaluate

the hypothesized orders of vocational interests through the RANDALL program (Tracey, 1997).

The underlying mechanism of the method is to compare the order predictions in a correlation

matrix with the hypothesized orders assumed in Holland’s theory that correlations between

adjacent interest types are greater than those between alternates types and in turn greater than

those between opposite types (Holland, 1997, p. 29). A correspondence index and p-value are

23

generated to indicate the degree to which the hypothesized orders are met and to test the null

hypothesis that the ordering is random, respectively (Rounds et al., 1992). The range of a

correspondence index is set between –1.00 to +1.00, where –1.00 indicates completely violation

and +1.00 means perfect model fit. One advantage of using correspondence index is that it

allows direct comparisons across studies and matrices (Rounds et al., 1992). Previous research

based on the Strong assessment suggests that U.S. samples and ethnic U.S. groups usually have a

correspondence index value larger than .70 (Kantamneni & Fouad, 2011; Kantamneni, 2014),

while international samples have lower correspondence index values (Rounds & Tracey, 1996).

This is also true in Long and Tracey’s (2006) meta-analysis that the mean correspondence index

value for Holland’s theory is .54 (SD = .22) across various Chinese samples from mainland

China, Hong Kong SAR, and Taiwan. As a rule of thumb, a p-value < .05 indicates the

hypothesis that random relabeling of six interest types in correlation matrices can be rejected for

the samples (Kantamneni & Fouad, 2011; Morgan & de Bruin, 2017). Therefore, the criteria of

correspondence index > .70 and p < .05 will be used to evaluate model fits in our sample.

However, the author would expect that the correspondence index value falls between .60 and .70.

The correlation matrices used for calculating correspondence index and p values for different

groups can be found in Table 5 to Table 9.

The circumplex covariance structure modeling (CCSM; Browne, 1992, for technical

specifications) is a promising confirmatory factor analysis strategy that “assesses the extent to

which the underlying structure of the correlation matrix is circumplex” (Fabringer, Visser, &

Brown, 1997, for non-technical narrative). The CCSM is conducted through the CircE package

(Grassi, Luccio, & di Blas, 2010) in RStudio (RStudio Team, 2016). This approach is

characterized by the calculation of parameter estimates (see the section of Overview of Holland’s

24

theory of vocational personalities) on each interest types and visualization of these estimates on

the circumference of a circle. Fit indexes (such as chi-square, RMSEA, and CFI) are also

provided to help researchers judge and evaluate the goodness of fit of the models. In the current

study, model fits were examined through several indexes including chi-square (χ2), root mean

square error of approximation (RMSEA), standardized root mean square residual (SRMR),

comparative fit index (CFI), and Tucker-Lewis Index (TLI). Since χ2, RMSEA, and SRMR are

sensitive to sample size and/or degree of freedom (χ2 is inclined to be significant for larger

sample; RMSEA and SRMR are biased for smaller df), CFI and TLI are incorporated as a means

of complementation. Two sets of combination rules, therefore, were used in this study to indicate

good model fits: (a) CFI ≥.95 and SRMR ≤ .08, and (b) TLI ≥ .95 and SRMR ≤ .08 (Hu &

Bentler, 1998; Hu & Bentler, 1999). Note that criteria aforementioned are not absolute indicators

of good or bad model fit and determinations of adequate fit should consider the synthetic

performance of all fit indices.

25

CHAPTER IV: TRANSLATION AND ADAPTATION PROCEDURES

The translation/back-translation technique is frequently seen in cross-cultural studies

where testing and assessments need to be adapted to a target language. However, the back-

translation procedure is not without limitations such as no evaluations on the target language

items (Harkness, 2003) and fewer emphases on commutations, naturalness, and

comprehensibility (Van de Vijver & Leung, 1997, p.39). Therefore, the forward translation

technique used in Long, Adams, and Tracey study (2005, for the Personal Globe Inventory) was

applied in the current work to translate and adapt the Strong into Simplified Chinese. This

method often contains two phases: (a) a group of bilingual individuals translates the assessment

into the target language individually, and (b) a team of reviewers judge the equivalence of the

source and target language versions and come up with a final version. As suggested by Harkness

(2003), the forward translation design is preferred if only one translation design is used.

Moreover, this method is also in line with the recommendation of the publisher of the Strong

assessment as well as best practices in the International Test Commission Guidelines (ITC,

2016).

The Translation and Review Committee

The committee approach (Geisinger & McCormick, 2013) that multiple bilingual

individuals translate the assessment from the original language to the target language is a

preferable way to generate more desirable translations. In addition, translators’ competencies, in

a large extent, can affect the quality of the translation (Goh & Yu, 2001). A qualified translator

should understand all meaning of the items in the original language and culture and decide the

most appropriate meaning in the target language and context (Kim, 2009). The translation and

review committee in this study consisted of four bilingual native Chinese people, two in China

26

and two in the U.S. All committee members were familiar with both cultures, fluent in American

English and Simplified Chinese, and holding a bachelor’s degree or higher in psychology or

related subjects. Two translators were appointed to translate the instructions and items into

Simplified Chinese. Specifically, one translator received a master’s degree in I/O Psychology

from an accredited Midwest university in the U.S. and is now working in China, whereas the

other is a current Ph.D. student in Management with psychology background at a Southeastern

university in the U.S. On the other side, two reviewers, including the author, were responsible to

review and reconcile the translations and generate the Chinese inventory used in this study. The

reviewer was appointed by the publisher of the Strong assessment who is the distributor of the

publisher’s other assessment products in China.

First Phase: Direct Translation

Two bilingual translators were asked to provide the translation of the Strong assessment,

respectively. In this stage, instructions and items were literally and directly translated into

Simplified Chines without any adaptation to achieve inferential equivalence between the original

language and the target language. Results of the comparison between two individual translation

work suggested that 105 items (36.1% of 291) reached a complete match. These items were

preliminarily considered without cultural adaptation.

Second Phase: Review and Reconciliation

For the remaining 186 items that were not identical, linguistic disagreements were settled

by two reviewers through (a) choosing a better translation from two versions, and (b) writing the

new translation based on the existing work, resulting in the match-rate increase by 39.5% (115

items) and 23.0% (47 items), as well as four items lacking linguistic equivalence. Specifically,

these four items are “bank teller,” “cashier in a bank,” “English composition,” and “prefer

27

working alone rather than on committees.” The author adopted three comparable translations

from Goh and Yu (2001) and made modifications for the last item. As a result, four items were

adapted to “bank teller/cashier,” “senior clerk in a bank,” “Chinese composition,” and “prefer

working alone rather than in teams,” respectively. Closer inspections suggested that “director of

religious education” and “religious leader (e.g., minister, monk, nun, priest, rabbi)” were unusual

occupations in Chinese culture, which might be culturally specific to the U.S. population. In

brief, 287 items achieved linguistic equivalence after modifications and four items were adjusted

specifically for Chinese culture.

28

CHAPTER V: RESULTS

Descriptive Statistics and Reliability

Six interest types represented by GOTs had satisfied Cronbach’s alpha reliability, with a

median alpha of .91 (Table 4). Almost all types except the Conventional had an alpha above .90,

ranging from .89 to .93. Means and standard deviations of GOTs for the sample and four

subgroups are also shown in Table 4. Based on GOT scores, inter-correlations between interest

types are yielded in Table 5 to 9 (see Table 5 for all respondents, Table 6 for males, Table 7 for

females, Table 8 for students, and Table 9 for employees) and used for RTHOR and CCSM.

Means, standard deviations, and reliability estimates of 30 BISs are presented in Table 10.

Cronbach’s alphas suggested acceptable to good reliability across BISs, ranging from .76 for

Office Management to .91 for Mathematics with a median of .85. The reliability for five PSSs

were also acceptable to good, ranging from .76 for Team Orientation to .90 for Learning

Environment with a median of .85 (Table 11).

Randomization Test

Results of RTHOR are presented in Table 12. The correspondence index values for the

sample and four subgroups were all significant with p-values ≤ .05. The overall sample had a

correspondence index value of .78 (> .70), indicating the circular structure had a satisfactory fit

to the Chinese sample in this study. As for four subgroups, however, correspondence index

values were lower than the U.S. benchmark of .70 but were all greater than .60. Specifically,

male participants (.64) had slightly higher correspondence index value than females (.61), and

the value for the student group (.69) was much closer to the benchmark than the employee group

(.61). In fact, correspondence index values in the current study were much better than these of

29

Chinese samples in Long and Tracey (2006, mean correspondence index = .54, SD = .22) and

international samples in Rounds and Tracey (1996, mean correspondence index = .48, SD = .18).

Circumplex Covariance Structure Modeling

Model fit statistics from the CCSM are shown in Table 13. Compared against the criteria

(CFI ≥.95, TLI ≥ .95, and SRMR ≤ .08), Holland’s circular ordering model (χ2 = 14.020, df = 3,

CFI = .988, TLI = .942, SRMR = .020) showed better fit to the overall sample than quasi-

circumplex models (for equal communality assumption, χ2 = 62.820, df = 8, CFI = .942, TLI =

.891, SRMR = .060; for equal angular location assumption, χ2 = 59.430, df = 8, CFI = .945, TLI

= .898, SRMR = .051) and circumplex model (χ2 = 110.840, df = 13, CFI = .896, TLI = .880,

SRMR = .073). The results of model fit were also replicated by all four subgroups (Table 13).

This is not surprising because the circular ordering model is the loosest one without any

parameters constrained, while the circumplex model is the most restrictive and is constrained by

RIASEC ordering and equal angular locations and communalities.

Therefore, a closer investigation was given to the circular ordering model across the

sample and four subgroups. As mentioned before (in the analysis section), one hallmark of the

circumplex covariance structure modeling is that it converts the correlation matrices into

comparable estimates of polar angles and communalities so that each interest types can be

geographically presented along the circumference of a circle. Table 14 presents the point

estimates and 95% confidence intervals for angular locations and communalities for six interest

types across the sample and subgroups, and the point estimates are further visualized in Figure 3

to Figure 7. Note that when examining these figures, the ordering of six interest types around the

circle can either be clockwise or counter-clockwise.

30

Obviously, the overall sample has an identical circular order of RIASEC that was

hypothesized by Holland (1973, 1985, 1997), although Realistic/Investigative and Social/Artistic

were much closer than any interest types (11˚ and 35˚, respectively, Figure 3). As for subgroups,

males and students fit the RIASEC order, while females and working adults were deviant from it.

More specifically, a closer angular locations can be found between Artistic and Social (5˚) in the

male group (Figure 4), and Realistic and Investigative (17˚) got much closer among students

(Figure 6). Both female and employee groups yielded the ordering of RASECI with Investigative

violate the assumed order between Realistic and Artistic. Furthermore, smaller polar angles were

found between Investigative and Realistic for these two groups (11˚ for the female group, 5˚ for

the employee group).

Comparisons between Two Cultures

Since the results of RTHOR and CCSM suggested that the overall Chinese sample, rather

than four subgroups, had a good fit of the RIASEC ordering theme, scores of GOT, BIS, and

PSS for the Chinese sample were then compared against the GRS (or normative group) with a

standardized mean of 50 and standard deviation of 10. An overall finding for scores across these

scales indicated that Chinese participants had a greater central tendency than the U.S. normative

group, in that the standard deviations of all scale scores in Table 3, 10 and, 11 were much lower

than 10. An investigation of GOT scores suggested that the Chinese sample had comparable

interests in Realistic, Investigative, and Enterprising, but higher interests in Artistic, Social, and

especially Conventional than the U.S. normative group. Comparisons of BIS scores also revealed

some interesting findings that the Chinese sample had lower mean scores on Athletics,

Mathematics, and Entrepreneurship, but relatively higher scores on Military, Sales, and Office

Management. These findings are further discussed in the following section.

31

CHAPTER VI: DISCUSSION

In this study, the 2004 Strong Interest Inventory was translated into Simplified Chinese

through several judgmental procedures including comparison, merging, and reconciliation.

Around 99% of items reached linguistic and inferential equivalence, and the remaining 1% was

replaced by comparable items that had similar theoretical meanings in Chinese culture. Several

items regarding religions were also found to lack cultural specificity in China. However, no

appropriate replacements were found to remedy this problem. Another few items related to

agricultural occupations might have had pejorative connotations. In brief, great efforts were

made to translate the instructions and items without altering originally underlying meanings and

ensure that non-English speakers in China can comprehend and respond to the items without

difficulties. This translation work is essential and meaningful to fill the void in the research and

practice of career assessment.

Reliability and validity of the Chinese Strong assessment were examined through

multiple statistical analyses including Cronbach’s alpha, RTHOR, and CCSM. Especially, the

latter two approaches were not heavily used by previous research on the applicability of

Holland’s model in China, which is a methodological advance that warrants the current study.

Results of RTHOR provided strong evidence that the overall sample fit the circular ordering

model well with a significant correspondence index value of .78. However, less support was

found for the RIASEC ordering of the four subgroup partitions (males, females, students, and

full-time employees) with correspondence index values ranging from .60 to .70. In addition, our

sample provided more favorable evidence than previous empirical and meta-analytic studies that

surveyed Chinese samples, which usually resulted in lower correspondence index values (≤ .60).

The CCSM results indicated that the overall sample (as well as the four subgroups) performed

32

better for the unconstrained loose circular model and worst for the restrictive circumplex model

judging from the model fit statistics. A further inspection of parameter estimates and geographic

presentations revealed that the overall sample and two subgroups (males and students) followed

the RIASEC ordering with some pairs of interest types tending to become closer regarding

angular locations. Cross-cultural comparisons on GOT, BIS, and PSS between U.S. normative

group and the current Chinese samples suggested comparable standardized scores with few

violated scales. To illustrate, the Chinese sample was inclined to have higher GOT scores on

Artistic, Social, and Conventional, and lower BIS scores on Athletics, Mathematics, and

Entrepreneurship.

Theoretical Implications

This study contributes to the body of vocational interest literature by translating an

authoritative U.S.-based interest inventory into Simplified Chinese and providing more up-to-

date insights into the RIASEC structure in Chinese culture by examining four specific Holland’s

models in a Chinese sample. Strong evidence was found for the circular ordering model in the

sample through two statistical approaches, RTHOR and CCSM. The inclusion of diverse groups

(students and working adults) with a large age range (15 to 50) remedied the drawback of sample

compositions in previous studies.

Moreover, existing empirical and meta-analytic research painting a controversial picture

of Holland’s theory in Chinese culture was challenged by the promising findings in the current

study that suggested strong support for the RIASEC ordering model in a diverse Chinese sample.

Although the sample was not adequately representative to generalize the conclusions to the

whole Chinese population, this study presented a more scientific inventory and several advanced

and effective statistical methods for researchers to replicate in the future.

33

Practical Implications

The practical implication of this study is that the Strong Interest Inventory can be a good

candidate used in the Chinese population for most of the ages. A comprehensive vocational

assessment has been developed and put into service by the Ministry of Education of the People’s

Republic of China for more than ten years (Ma, 2003), aiming at providing high school students

with scientific and valid vocational information regarding their interests as well as aptitude.

However, the set of assessments is not applicable for college students, working adults, and

people seeking job opportunities. On the other hand, although Personal Globe Inventory (PGI)

and Self-Directed Search (SDS) have received more attention than the Strong Interest Inventory

(Donnay et al., 2004) in mainland China, the transportability and generalizability of the Chinese

forms of these inventories were still questionable and, to the author’s knowledge, PGI and SDS

have not been widely used and commercialized in the Chinese market. Therefore, the promising

findings in this study may potentially increase the assessment tools for career counselors in

China in the future.

Limitations and Future Research

The present study is not without limitations and should be addressed by further research.

First, several items off occupations and activities in China are not as usual as them in the U.S.

such as “spiritual leader.” The translation and review procedure have failed to come up with

appropriate substitutions with equivalent underlying meanings, which undoubtedly decrease the

cross-cultural validity of the Strong assessment given the notion of culture-free and bias-free

assessments (Geisinger & McCormick, 2013). Future research should recruit linguists into the

committee to attend to this limitation.

34

Second, as noticed in the data-cleaning section, the response pattern and item

endorsement in the Chinese sample is different from the U.S. normative group, which makes it

difficult to identify bad cases as well as comparing the Chinese respondents with the appropriate

culturally specific norm. An investigation of response percentages tells that the Chinese sample

in more inclined to have middle category endorsements (i.e., indifferent) than extreme ones (e.g.,

strongly like) due to the unique Chinese culture that values modesty and humility. The addition

research is encouraged to put some emphasis on the effect of cultural factors on interest item

responses.

Third, as repeatedly mentioned by previous studies in 1994 SII, additional studies using

the Chinese translation of the current Strong assessment are expected to replicate the favorable

findings in this study. Since the snowball sampling used in this study is a non-probability

sampling technique where existing respondents are asked to recruit future participants from their

acquaintances, “community bias” may generate from the potentially homogeneous samples

recruited through this approach albeit it is useful to access to hard-to-reach populations

(Heckathorn, 2011). Therefore, future research is encouraged to use probability sampling

methods such as stratified random sampling and systematic random sampling to collect

representative samples resembling the population composition of China and use them for cross-

cultural validation and norm development.

35

CHAPTER VII: CONCLUSION

In this study, the author (together with the committee) translated the 2004 Strong Interest

Inventory into Simplified Chinese and tested four forms of Holland’s model on a diverse Chinese

sample. This study extends existing vocational literature by adding the knowledge of the cross-

cultural validity of the latest Strong assessment in Chinese culture. In conclusion, the findings

suggests that the Chinese sample and two subgroups (males and students) have the identical

RIASEC ordering hypothesized by Holland and the Strong Interest Inventory is reliable and

valid and can be a promising vocational assessment tool used in China.

36

References

Anderson, M. Z., Tracey, T. G., & Rounds, J. (1997). Examining the invariance of Holland's

vocational interest model across gender. Journal of Vocational Behavior, 50(3), 349-364.

http://dx.doi.org/10.1006/jvbe.1996.1550

Armstrong, P. I., Hubert, L., & Rounds, J. (2003). Circular unidimensional scaling: A new look

at group differences in interest structure. Journal of Counseling Psychology, 50(3), 297-

308. http://dx.doi.org/10.1037/0022-0167.50.3.297

Browne, M. W. (1992). Circumplex models for correlation matrices. Psychometrika, 57(4), 469-

497. http://dx.doi.org/10.1007/BF02294416

Bryant, F. B., & Yarnold, P. R. (1995). Principal-components analysis and exploratory and

confirmatory factor analysis. In L. G. Grimm, P. R. Yarnold, L. G. Grimm, P. R. Yarnold

(Eds.) , Reading and understanding multivariate statistics (pp. 99-136). Washington, DC,

US: American Psychological Association.

Bullock, E. E., Andrews, L., Braud, J., & Reardon, R. C. (2009). Holland’s theory in an

international context: Applicability of RIASEC structure and assessments. Career

Planning and Adult Development Journal, 25(4), 29-58.

Comfrey, A. L., & Lee,H. B. (1992). A first course in factor analysis. Hillsdale, NJ: Lawrence

Erlbaum Associates.

Darcy, M. A., & Tracey, T. G. (2007). Circumplex structure of Holland's RIASEC Interests

across gender and time. Journal of Counseling Psychology, 54(1), 17-31.

http://dx.doi.org/10.1037/0022-0167.54.1.17

37

Day, S. X., & Rounds, J. (1998). Universality of vocational interest structure among racial and

ethnic minorities. American Psychologist, 53(7), 728-736.

http://dx.doi.org/10.1037/0003-066X.53.7.728

Day, S. X., Rounds, J., & Swaney, K. (1998). The structure of vocational interests for diverse

racial-ethnic groups. Psychological Science, 9(1), 40-44. http://dx.doi.org/10.1111/1467-

9280.00007

Donnay, D. A. C., Morris, M. L., Schaubhut, N. A., & Thompson, R. C. (2004). Strong Interest

Inventory manual: Research, development, and strategies for interpretation. Palo Alto,

CA: Consulting Psychologists Press Incorporated.

Fabrigar, L. R., Visser, P. S., & Browne, M. W. (1997). Conceptual and methodological issues in

testing the circumplex structure of data in personality and social psychology. Personality

and Social Psychology Review, 1(3), 184-203.

http://dx.doi.org/10.1207/s15327957pspr0103_1

Fan, W., & Leong, F. L. (2016). Introduction to the special issue: Career development and

intervention in Chinese contexts. Career Development Quarterly, 64(3), 192-202.

http://dx.doi.org/10.1002/cdq.12054

Farth, J. L., Leong, F. T., & Law, K. S. (1998). Cross-cultural validity of Holland's model in

Hong Kong. Journal of vocational behavior, 52(3), 425-440.

https://doi.org/10.1006/jvbe.1997.1631

Fouad, N. A. (1993). Cross-cultural vocational assessment. The Career Development

Quarterly, 42(1), 4-13. https://doi.org/10.1002/j.2161-0045.1993.tb00240.x

38

Fouad, N. A. (2002). Cross-cultural differences in vocational interests: Between-group

differences on the Strong Interest Inventory. Journal of Counseling Psychology, 49(3),

283-289. http://dx.doi.org/10.1037/0022-0167.49.3.282

Fouad, N. A., Harmon, L. W., & Borgen, F. H. (1997). Structure of interests in employed male

and female members of US racial-ethnic minority and nonminority groups. Journal of

Counseling Psychology, 44(4), 339-345. http://dx.doi.org/10.1037/0022-0167.44.4.339

Gasser, C. E., Larson, L. M., & Borgen, F. H. (2007). Concurrent validity of the 2005 Strong

Interest Inventory: An examination of gender and major field of study. Journal of Career

Assessment, 15(1), 23-43. http://dx.doi.org/10.1177/1069072706294516

Ge, S., Yu, J., & Wang, S. (1996). The Chinese translation of the Strong Interest Inventory (SII).

Journal of Nanjing Normal University (Social Science Edition), (3), 46-49. (in Chinese)

[葛树人，余嘉元，万书玉. 斯特朗职业兴趣调查表（SII）的中文翻译. 南京师范大

学学报（社会科学版）, 1996, (3): 46-49.] Retrieved from:

http://kns.cnki.net/KCMS/detail/detail.aspx?dbcode=CJFQ&dbname=CJFD9697&filena

me=NJSS603.009&v=MTI0MDdkWmVkcEZpN2hWdz09S3lmWWZiVzRIYy9NcjRZc

UY1NE9mZ2c1emhBVTRqaDRPWDZUckgwM2ViT1RSYjY=

Geisinger, K. F., & McCormick, C. (2013). Testing and assessment in cross-cultural psychology.

In J. R. Graham, J. A. Naglieri, I. B. Weiner, J. R. Graham, J. A. Naglieri, I. B. Weiner

(Eds.) , Handbook of psychology: Assessment psychology (pp. 114-139). Hoboken, NJ,

US: John Wiley & Sons Inc.

Glidden-Tracey, C., & Greenwood, A. K. (1997). A validation study of the Spanish Self-

Directed Search using back-translation procedures. Journal of Career Assessment, 5(1),

105-113. http://dx.doi.org/10.1177/106907279700500107

39

Goh, D. S. (2001). Translation and validation of the Chinese form of the Strong Interest

Inventory. Applied Psychology: An International Review, 50(2), 252-268.

http://dx.doi.org/10.1111/1464-0597.00057

Goh, D. S., Lee, J. A., & Yu, J. (2004). Factor structure of the Strong Interest Inventory with a

Chinese high school sample. Journal of Psychology, 138(2), 171-183.

http://dx.doi.org/10.3200/JRLP.138.2.171-184

Goodman, L. A. (1961). Snowball sampling. Annals of Mathematical Statistics, 32, 148-170.

http://dx.doi.org/doi:10.1214/aoms/1177705148

Gorusch, R. L. (1983). Factor Analysis (2nd ed.). Hillsdale, NJ: Lawrence Erlbaum Associates.

Grassi, M., Luccio, R., & di Blas, L. (2010). CircE: An R implementation of Browne's circular

stochastic process model. Behavior Research Methods, 42(1), 55-73.

http://dx.doi.org/10.3758/BRM.42.1.55

Guttman, L. R. (1954). A new approach to factor analysis: The radex. In P. F. Lazarsfeld (Ed.),

Mathematical thinking in the social sciences (pp. 258-348). New York: Columbia

University Press.

Hansen, J. C., & Fouad, N. A. (1984). Translation and validation of the Spanish form of the

Strong-Campbell Interest Inventory. Measurement & Evaluation in Guidance, 16(4),

192-197.

Hao, D., Sun, V. J., & Yuen, M. (2015). Towards a model of career guidance and counseling for

university students in China. International Journal for the Advancement of Counselling,

37(2), 155-167. http://dx.doi.org/10.1007/s10447-015-9234-7

Harkness, J. A. (2003). Questionnaire translation. In J. A. Harkness, F. J. R. van de Vijver, and

P.Ph. Mohler (Eds.), Cross-cultural survey methods (pp. 35-56). Hoboken, NJ: Wiley.

40

Harmon, L. W., Hansen, J-I., Borgen, F. H., & Hammer, A. L. (1994). Strong Interest Inventory:

Applications and technical guide. Palo Alto, CA: Consulting Psychologist Press.

Heckathorn, D. D. (2011). Snowball versus respondent-driven sampling. Sociological

Methodology, 41(1), 355-366. http://dx.doi.org/10.1111/j.1467-9531.2011.01244.x

Herk, N. A., & Thompson, R. C. (2011). International Technical Brief for the Strong Interest

Inventory Assessment. Mountain View, CA: CPP, Inc.

Hogan, R. (1983). A socioanalytic theory of personality. In M. M. Page (Ed.), Nebraska

symposium on motivation 1982. Personality: Current theory and research (pp. 55-89).

Lincoln, NE: University of Nebraska Press.

Holland, J. L. (1973). Making vocational choices: A theory of vocational personalities & work

environments. Englewood Cliffs, NJ: Printice Hall.


environments (2nd ed.). Englewood Cliffs, NJ: Printice Hall.


environments (3rd ed.). Lutz, FL: Psychological Assessment Resources.

Hu, L., & Bentler, P. M. (1998). Fit indices in covariance structure modeling: Sensitivity to

underparamaterized model misspecification. Psychological Methods, 3, 424-453.

http://dx.doi.org/10.1037/1082-989X.3.4.424

Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure analysis:

Conventional criteria versus new alternatives. Structural Equation Modeling, 6(1), 1-55.

http://dx.doi.org/10.1080/10705519909540118

Hubert, L., & Arabie, P. (1987). Evaluating order hypotheses within proximity matrices.

Psychological Bulletin, 102(1), 172-178. http://dx.doi.org/10.1037/0033-2909.102.1.172

41

International Test Commission. (2016). The ITC Guidelines for Translating and Adapting Tests

(Second edition). [www.InTestCom.org]

Kantamneni, N. (2014). Vocational interest structures for Asian Americans, Middle-Eastern

Americans and Native Americans on the 2005 Strong Interest Inventory. Journal of

Vocational Behavior, 84(2), 133-141. http://dx.doi.org/10.1016/j.jvb.2013.11.003

Kantamneni, N., & Fouad, N. (2011). Structure of vocational interests for diverse groups on the

2005 strong interest inventory. Journal of Vocational Behavior, 78(2), 193-201.

http://dx.doi.org/10.1016/j.jvb.2010.06.003

Kim, M. (2009). Meaning-Oriented Assessment of Translations: SFL and Its Application to

Formative Assessment. In C. V. Angelelli, H. E. Jacobson (Eds.), Testing and assessment

in translation and interpreting studies: A call for dialogue between research and practice

(pp. 123-157). Amsterdam, Netherlands: Benjamins.

http://dx.doi.org/10.1075/ata.xiv.08kim

Leong, F. L., & Brown, M. T. (1995). Theoretical issues in cross-cultural career development:

Cultural validity and cultural specificity. In W. B. Walsh, S. H. Osipow, W. B. Walsh, S.

H. Osipow (Eds.), Handbook of vocational psychology: Theory, research, and practice

(pp. 143-180). Hillsdale, NJ, US: Lawrence Erlbaum Associates, Inc.

Leong, F. L., Hartung, P. J., & Pearce, M. (2014). Work and career development: Theory and

research. In F. L. Leong, L. Comas-Díaz, G. C. Nagayama Hall, V. C. McLoyd, J. E.

Trimble, F. L. Leong, ... J. E. Trimble (Eds.), APA handbook of multicultural psychology,

Vol. 1: Theory and research (pp. 451-469). Washington, DC, US: American

Psychological Association. http://dx.doi.org/10.1037/14189-024

42

Leong, F., Austin, J., Sekaran, U., & Komarraju, M. (1998). An evaluation of the cross-cultural

validity of Holland's theory: Career choices by workers in India. Journal of Vocational

Behavior, 52(3), 441-455. http://dx.doi.org/10.1006/jvbe.1997.1637

Long, L., & Tracey, T. J. (2006). Structure of RIASEC scores in China: A structural meta-

analysis. Journal of Vocational Behavior, 68(1), 39-51.


Long, L., Adams, R. S., & Tracey, T. J. G. (2005). Generalizability of interest structure to China:

Application of the Personal Globe Inventory. Journal of Vocational Behavior, 66, 66-80.


Lowman, R. L., & Carson, A. D. (2013). Conceptualization and assessment of interests. In J. R.

Graham, J. A. Naglieri, I. B. Weiner, J. R. Graham, J. A. Naglieri, I. B. Weiner (Eds.) ,

Handbook of Psychology: Assessment Psychology (pp. 534-557). Hoboken, NJ, US: John

Wiley & Sons Inc.

Ma, S. (2003). Development report for Guidance Assessment of Further Education and

Employment. Psychological Exploration, 23(3), 60-62.

Marsella, A. J., & Leong, F. L. (1995). Cross-cultural issues in personality and career

assessment. Journal of Career Assessment, 3(2), 202-218.

http://dx.doi.org/10.1177/106907279500300207

Morgan, B., & de Bruin, G. P. (2017). Structural validity of Holland’s circumplex model of

vocational personality types in Africa. Journal of Career Assessment, 25(1), 1-17.

http://dx.doi.org/10.1177/1069072717692747

43

Morris, M. L. (2016). Vocational Interests in the United States: Sex, Age, Ethnicity, and Year

Effects. Journal of Counseling Psychology, 63(5), 604-615.

http://dx.doi.org/10.1037/cou0000164

Nye, C. D., Su, R., Rounds, J., & Drasgow, F. (2017). Interest congruence and performance:

Revisiting recent meta-analytic findings. Journal of Vocational Behavior, 98, 138-151.


Rounds, J., & Tracey, T. J. (1996). Cross-cultural structural equivalence of RIASEC models and

measures. Journal of Counseling Psychology, (3), 310-329.

http://dx.doi.org/10.1037/0022-0167.43.3.310

Rounds, J., Tracey, T. J., & Hubert, L. (1992). Methods for evaluating vocational interest

structural hypotheses. Journal of Vocational Behavior, 40(2), 239-259.

http://dx.doi.org/10.1016/0001-8791(92)90073-9

RStudio Team (2016). RStudio: Integrated Development for R (Version 1.0.136) [Software].

Available from: http://www.rstudio.com/.

Sodano, S. M. (2015). Meaning, measurement, and assessment of vocational interests for career

intervention. In P. J. Hartung, M. L. Savickas, W. B. Walsh, P. J. Hartung, M. L.

Savickas, W. B. Walsh (Eds.) , APA handbook of career intervention, Volume 1:

Foundations (pp. 281-301). Washington, DC, US: American Psychological Association.

http://dx.doi.org/10.1037/14438-016

Su, R., Rounds, J., & Armstrong, P. I. (2009). Men and things, women and people: a meta-

analysis of sex differences in interests. Psychological Bulletin, 135(6), 859-884.

http://dx.doi.org/10.1037/a0017364

44

Subich, L. M. (2005). Career Assessment with Culturally Diverse Individuals. In W. B. Walsh,

M. L. Savickas, W. B. Walsh, M. L. Savickas (Eds.), Handbook of vocational

psychology: Theory, research, and practice, 3rd ed (pp. 397-421). Mahwah, NJ, US:

Lawrence Erlbaum Associates Publishers.

Tang, M. (2001). Investigation of the structure of vocational interests of Chinese college

students. Journal of Career Assessment, 9(4), 365-380.

http://dx.doi.org/10.1177/106907270100900404

Tang, M. (2009). Examining the application of Holland's theory to vocational interests and

choices of Chinese college students. Journal of Career Assessment, 17(1), 86-98.

http://dx.doi.org/10.1177/1069072708325743

Tracey, T. G. (2000). Analysis of circumplex models. In H. A. Tinsley, S. D. Brown, H. A.

Tinsley, S. D. Brown (Eds.) , Handbook of applied multivariate statistics and

mathematical modeling (pp. 641-664). San Diego, CA, US: Academic Press.

http://dx.doi.org/10.1016/B978-012691360-6/50023-9

Tracey, T. J. (1997). RANDALL: A Microsoft FORTRAN program for a randomization test of

hypothesized order relations. Educational and Psychological Measurement, 57(1), 164-

168. http://dx.doi.org/10.1177/0013164497057001012

Tracey, T. J., & Robbins, S. B. (2005). Stability of interests across ethnicity and gender: A

longitudinal examination of grades 8 through 12. Journal of Vocational Behavior, 67(3),

335-364. http://dx.doi.org/10.1016/j.jvb.2004.11.003

Tracey, T. J., & Rounds, J. (1993). Evaluating Holland's and Gati's vocational-interest models: a

structural meta-analysis. Psychological Bulletin, 113(2), 229-246.

http://dx.doi.org/10.1037/0033-2909.113.2.229

45

Van de Vijver, F. d., & Kwok, L. (2000). Methodological issues in psychological research on

culture. Journal of Cross-Cultural Psychology, 31(1), 33-51.

http://dx.doi.org/10.1177/0022022100031001004

Van de Vijver, F. J. R., & Leung, K. (1997). Methods and data analysis for cross-cultural

research. Thousand Oaks, CA: Sage.

Van de Vijver, F. R. (2015). Methodological aspects of cross-cultural research. In M. J. Gelfand,

C. Chiu, Y. Hong, M. J. Gelfand, C. Chiu, Y. Hong (Eds.) , Handbook of advances in

culture and psychology, Vol. 5 (pp. 101-160). New York, NY, US: Oxford University

Press. http://dx.doi.org/10.1093/acprof:oso/9780190218966.003.0003

Wang, L., & Shen, W. (1997). A comprehensive introduction to The Strong-Campbell Interest

Inventory (SCII). Journal of Developments in Psychology, 5(2), 28-33. (in Chinese) [王

垒，沈伟. 斯特朗-坎贝尔兴趣量表（SCII）的综合介绍. 心理学动态, 1997, 5(2): 28-

33.] Retrieved from:

http://kns.cnki.net/KCMS/detail/detail.aspx?dbcode=CJFQ&dbname=CJFD9697&filena

me=XLXD702.005&v=MDIxNTZVQT09UFNIVGFyUzRITS9NcjRvcUY1NE9mZ2c1e

mhBVTRqaDRPWDZUckgwM2ViT1RSYjZkWmVkcEZTM20=

Wang, Y., Xue, X., Li, Y., & Zhang, X. (2016). Empirical validity of the Personal Globe

Inventory-Short Chinese form. Chinese Human Resource Development, (11), 66-71. (in

Chinese) [王宇中，薛雪，李亚欧，张晓宇. 个人球形职业兴趣量表简版中国版

（PGI-SC）的实证效度：基于 20 个职业领域的调查资料. 中国人力资源开发, (11),

66-71.] Retrieved from:

http://kns.cnki.net/KCMS/detail/detail.aspx?dbcode=CJFQ&dbname=CJFDLAST2016&

46

filename=ZRZK201611011&v=MzE2MDdmWStSdEZ5L25VYnZBUHovUlpiRzRIOW

ZOcm85RVpZUjhlWDFMdXhZUzdEaDFUM3FUcldNMUZyQ1VSTDI=

Westermeyer, J. (1987). Cultural factors in clinical assessment. Journal of Consulting and

Clinical Psychology, 55(4), 471-477. http://dx.doi.org/10.1037/0022-006X.55.4.471

Yan, L. (2008). Localization of career counseling in china: Achievements and challenges. Career

Planning and Adult Development Journal, 24(4), 39-43.

Yang, W., Lance, C. E., & Hui, H. C. (2006). Psychometric properties of the Chinese Self-

Directed Search (1994 edition). Journal of Vocational Behavior, 68(3), 560-576.


Yang, W., Stokes, G. S., & Hui, C. H. (2005). Cross-cultural validation of Holland’s interest

structure in Chinese population. Journal of Vocational Behavior, 67(3), 79-396.


Zhang, Y., Wei, Q., Li, H., Wang, Y. & Tracey, J. G. T. (2015). Reliability and validity of the

Personal Globe Inventory-short in China. Chinese Journal of Applied Psychology, 21(1),

39-48. (in Chinese) [张宇，魏青，李红，王宇中，Terence J. G. Tracey. 个人球形职

业兴趣量表简版（PGI-S）中文版的信效度检验. 应用心理学, 2015, 21(1), 39-48.]

Retrieved from:

http://kns.cnki.net/KCMS/detail/detail.aspx?dbcode=CJFQ&dbname=CJFDLAST2015&

filename=YXNX201501005&v=MDg5NzExVDNxVHJXTTFGckNVUkwyZlkrUnRGe

S9uVWI3QlBEWEZkckc0SDlUTXJvOUZZWVI4ZVgxTHV4WVM3RGg=

47

Appendix A: Tables

Table 1

Summary of Research on Testing Holland’s Model in China

Reference Instrument

(Language)

Sample Analyses Gender

Difference

Major Findings

Goh & Yu (2001) SII-1994

(Chinese)

124 Chinese college students in

Southeast China and 40

bilingual Chinese college

student in the U.S.

• Correlation

• T-test

• Profile analysis

• EFA

Yes • Six-factor model was yielded

• Basic Interest Scales did not resemble to the

original classification

• “Conventional” was changed to “Public”

Tang (2001) SII-1994

(Chinese)


Northeastern China, mean age

= 21.59, age range = 18 to 24

• MDS

• EFA

• Discriminant

Analysis

Yes • Support for model fit

• No support for calculus hypothesis

• Basic Interest Scales did not resemble to the

original classification

Goh, Lee, & Yu

(2004)

SII-1994

(Chinese)

247 Chinese high school

students in Nanjing, age range

= 15 to 18

• CFA

• Correlation

• EFA

N/A • No support for model fit

• No support for calculus hypothesis

• A three-factor model was better

Yang, Stokes, &

Hui (2005)

SDS-1994

(Chinese)

528 Chinese from Hong Kong

SAR and 325 Chinese from

mainland China, age range = 18

to 50

• CFA

• Randomization

Test

Yes • No support for circumplex model across

geographic and gender subgroups

• Mixed support for circular model across

geographic and gender subgroups

Yang, Lance, &

Hui (2006)

SDS-1994

(Chines)

528 Chinese from Hong Kong

SAR and 150 Chinese from

mainland China, age range = 18

to 50

• CFA

• MTMM

No • Full support across people from Hong Kong

SAR and mainland China as well as different

gender

Tang (2009) SDS-1994

(Chinese)


Northeastern China, mean age

= 21.6

• MDS

• Congruence

Scores

Yes • Mixed support

• Identical ordering but different distances of

RIASEC for males

• Different ordering but identical distances of

RIASEC for females

Note. MDS = multidimensional scaling; EFA = exploratory factor analysis; CFA = confirmatory factor analysis

48

Table 2

Suggestions from Studies on Testing Holland’s Model in China

Reference Suggestions Category

Goh & Yu

(2001) • Future research with larger samples is needed to cross-validate these

findings as well as to clarify some unresolved issues (such as

incomparable items)

Sample

Tang (2001) • Future studies might explore further the issues of universality of vocational

structure by incorporating more samples from various cultures

• Using a multifaceted approach, longitudinal method, and cross-validation

studies will also advance the research about vocational interests

Sample

Method

Goh, Lee, & Yu

(2004) • One suggestion is to administer the SII-Chinese to a large standardization

sample and use those data to determine its internal structure

Sample

Yang, Stokes,

& Hui (2005) • To be representative of the general population, future studies should

attempt other sampling methods

• Future cross-cultural validation of Holland’s interest structure can

similarly acknowledge the existence of moderating variables so as to make

the theory more useful and to more adequately represent the reality

Sample

Moderators

Yang, Lance, &

Hui (2006) • Further research should examine the Chinese SDS more closely and

culturally inappropriate items should be adapted to the local context

Items

Tang (2009) • To further examine the application of Holland’s theory in cross-cultural

settings, a larger sample with national representation and cross-sectional

validation studies are necessary

• Further studies should also explore what factors other than demographics

would influence congruence between individuals’ interests and career

choices

Sample

Moderators

49

Table 3

Demographic Information of Participants

Number Percent (%)

Gender

Male 135 37.1

Female 229 62.9

Education Level

Some high school 48 13.2

High-school diploma 30 8.2

Trade/Technical Training 2 0.5

Some college (no degree) 19 5.2

Associate/Community college degree 33 9.1

Bachelor’s degree 177 48.6

Master’s degree 50 13.7

Professional degree 1 0.3

Doctorate 4 1.1

Present Status

Working full-time 127 34.9

Working part-time 6 1.6

Not working for income 2 0.5

Retired 1 0.3

Enrolled as a full-time student 191 52.5

Seeking for a job 29 8.0

None of the above 8 2.2

Total 364 100.0

50

Table 4

GOT Reliability Statistics

Type Cronbach’s α Overall

M (SD)

Males

M (SD)

Females

M (SD)

Students

M (SD)

Employees

M (SD)

R .906 51.52 (7.85) 54.50 (7.62) 49.76 (7.47) 50.18 (8.09) 53.15 (7.14)

I .923 50.07 (8.34) 52.05 (8.43) 48.91 (8.08) 49.01 (8.86) 51.44 (7.43)

A .929 53.93 (7.31) 52.16 (6.86) 54.98 (7.39) 52.57 (7.09) 55.32 (7.19)

S .909 53.36 (7.64) 52.74 (7.23) 53.73 (7.87) 52.09 (7.94) 54.74 (7.24)

E .909 51.70 (8.27) 51.63 (7.82) 51.75 (8.54) 50.62 (8.40) 52.75 (7.67)

C .889 57.04 (8.05) 57.98 (7.99) 56.49 (8.06) 56.94 (8.03) 56.86 (8.31)

Note. N (overall) = 364; N (male) = 135; N (female) =229; N (student) = 191; N (employees) = 127; R = Realistic; I

= Investigative; A = Artistic; S = Social; E = Enterprising; C = Conventional.

51

Table 5

Correlation Matrix for the Overall Sample (N = 364)

R I A S E C

Realistic (R) 1.000

Investigative (I) 0.728 1.000

Artistic (A) 0.393 0.408 1.000

Social (S) 0.466 0.485 0.555 1.000

Enterprising (E) 0.368 0.263 0.338 0.610 1.000

Conventional (C) 0.492 0.474 0.321 0.531 0.603 1.000

52

Table 6

Correlation Matrix for the Male Group (N = 135)

R I A S E C

Realistic (R) 1.000


Artistic (A) 0.298 0.278 1.000

Social (S) 0.465 0.493 0.596 1.000

Enterprising (E) 0.376 0.225 0.403 0.569 1.000

Conventional (C) 0.466 0.372 0.223 0.525 0.591 1.000

53

Table 7

Correlation Matrix for the Female Group (N = 229)

R I A S E C

Realistic (R) 1.000


Artistic (A) 0.577 0.561 1.000

Social (S) 0.531 0.515 0.533 1.000

Enterprising (E) 0.394 0.294 0.313 0.631 1.000

Conventional (C) 0.502 0.525 0.412 0.549 0.615 1.000

54

Table 8

Correlation Matrix for the Student Group (N = 191)

R I A S E C

Realistic (R) 1.000


Artistic (A) 0.258 0.298 1.000

Social (S) 0.413 0.467 0.490 1.000

Enterprising (E) 0.337 0.212 0.248 0.600 1.000

Conventional (C) 0.456 0.455 0.217 0.512 0.595 1.000

55

Table 9

Correlation Matrix for the Employee Group (N = 127)

R I A S E C

Realistic (R) 1.000


Artistic (A) 0.507 0.531 1.000

Social (S) 0.471 0.461 0.561 1.000

Enterprising (E) 0.393 0.365 0.368 0.620 1.000

Conventional (C) 0.573 0.597 0.489 0.697 0.670 1.000

56

Table 10

BIS Reliability Statistics

Basic Interest Scale Cronbach’s α Overall

M (SD)

Males

M (SD)

Females

M (SD)

Realistic

Mechanics & Construction .861 51.22 (7.80) 54.01 (7.90) 49.57 (7.27)

Computer Hardware & Electronics .905 51.17 (7.95) 54.76 (7.71) 49.05 (7.32)

Military .863 55.62 (8.79) 58.37 (8.79) 53.99 (8.40)

Protective Services .783 52.38 (7.76) 53.37 (7.36) 51.80 (7.94)

Nature & Agriculture .863 50.67 (7.11) 50.51 (6.76) 50.76 (7.32)

Athletics .870 49.87 (7.31) 51.88 (7.21) 48.68 (7.11)

Investigative

Science .853 50.70 (8.20) 52.63 (8.54) 49.56 (7.78)

Research .857 51.04 (9.12) 53.11 (8.92) 49.81 (9.03)

Medical Science .826 51.42 (8.02) 51.60 (7.81) 51.32 (8.15)

Mathematics .908 49.73 (8.11) 52.04 (8.21) 48.37 (7.76)

Artistic

Visual Arts & Design .863 54.02 (7.71) 52.49 (7.47) 54.93 (7.71)

Performing Arts .855 52.49 (7.89) 50.17 (7.14) 53.86 (8.00)

Writing & Mass Communication .862 51.25 (7.44) 50.01 (7.22) 51.97 (7.49)

Culinary Arts .832 52.12 (7.60) 50.63 (7.30) 52.99 (7.65)

Social

Counselling & Helping .779 52.90 (7.05) 52.51 (6.79) 53.12 (7.21)

Teaching & Education .849 53.30 (7.72) 52.43 (7.82) 53.82 (7.64)

Humans Resources & Training .827 52.35 (8.10) 51.75 (7.57) 52.70 (8.39)

Social Sciences .785 51.55 (7.82) 52.30 (7.42) 51.10 (8.03)

Religion & Spirituality .856 50.79 (7.13) 51.17 (7.48) 50.57 (6.93)

Healthcare Services .844 51.60 (7.94) 51.14 (7.81) 51.87 (8.03)

Enterprising

Marketing & Advertising .827 50.48 (7.98) 50.12 (8.08) 50.69 (7.93)

Sales .877 57.11 (8.64) 57.54 (8.04) 56.87 (8.97)

57

Management .803 53.61 (8.29) 53.83 (8.07) 53.49 (8.43)

Entrepreneurship .792 47.04 (8.22) 47.66 (8.38) 46.68 (8.12)

Politics & Public Speaking .839 51.70 (7.01) 52.65 (6.83) 51.14 (7.07)

Law .871 51.46 (7.20) 51.20 (6.87) 51.61 (7.40)

Conventional

Office Management .755 56.78 (7.33) 55.90 (7.43) 57.30 (7.23)

Taxes & Accounting .822 53.02 (7.70) 54.27 (7.95) 52.29 (7.47)

Programming & Information Systems .849 50.00 (7.78) 52.38 (7.60) 48.59 (7.56)

Finance & Investing .834 52.58 (8.07) 53.92 (7.84) 51.79 (8.12)

Note. N (overall) = 364, N (male) = 135, N (female) =229.

58

Table 11

PSS Reliability Statistics

Personal Style Scale Cronbach’s α Overall

M (SD)

Males

M (SD)

Females

M (SD)

Work Style .866 51.55 (6.34) 49.17 (6.47) 52.95 (5.84)

Learning Environment .901 49.34 (6.39) 48.91 (6.80) 49.59 (6.13)

Leadership Style .852 50.00 (8.20) 50.59 (8.12) 49.65 (8.25)

Risk Taking .772 49.77 (7.83) 51.76 (7.32) 48.60 (7.90)

Team Orientation .761 50.29 (8.86) 50.56 (8.84) 50.13 (8.88)

Note. N (overall) = 364, N (male) = 135, N (female) =229.

59

Table 12

Results of Randomization Test of Hypothesized Ordering Relations

Predictions

Group N CI p Met Tied Not Met

Overall 364 .78 .017* 64 0 8

Male 135 .64 .017* 59 0 13

Female 229 .61 .033* 58 0 14

Students 191 .69 .017* 61 0 11

Employees 127 .61 .017* 58 0 14

Note. CI = correspondence index.

* p < .05.

60

Table 13

Model Fit Statistics for Circumplex Covariance Structure Modeling

Group Testing Model χ2 df RMSEA

[90% CI] SRMR CFI TLI

Overall Circular Ordering 14.020 3 .101 [.052, .156] 0.020 0.988 0.942

Quasi Equal Comm. 62.820 8 .137 [.107, .170] 0.060 0.942 0.891

Quasi Equal Ang. 59.430 8 .133 [.103, .166] 0.051 0.945 0.898

Circumplex 110.840 13 .144 [.120, .169] 0.073 0.896 0.880

Males Circular Ordering * 9.340 3 .126 [.039, .221] 0.029 0.979 0.934

Quasi Equal Comm. 28.730 8 .155 [.059, .307] 0.072 0.931 0.871

Quasi Equal Ang. 29.800 8 .143 [.090, .199] 0.061 0.928 0.865

Circumplex 50.050 13 .146 [.104, .190] 0.088 0.877 0.858

Females Circular Ordering * 14.350 3 .107 [.051, .169] 0.023 0.984 0.944

Quasi Equal Comm. 49.180 8 .150 [.112, .192] 0.055 0.940 0.888

Quasi Equal Ang. 61.910 8 .237 [.144, .363] 0.067 0.922 0.853

Circumplex 86.790 13 .158 [.127, .190] 0.078 0.893 0.876

Students Circular Ordering * 10.180 3 .090 [.020, .161] 0.025 0.984 0.947

Quasi Equal Comm. 48.950 8 .215 [.119, .353] 0.071 0.907 0.826

Quasi Equal Ang. 34.750 8 .133 [.089, .179] 0.053 0.939 0.886

Circumplex 81.940 13 .167 [.134, .203] 0.096 0.844 0.820

Employees Circular Ordering 2.650 3 .000 [.001, .143] 0.016 1.000 1.005

Quasi Equal Comm. 16.040 8 .089 [.017, .153] 0.049 0.979 0.961

Quasi Equal Ang. 29.250 8 .145 [.091, .203] 0.054 0.945 0.897

Circumplex 42.410 13 .134 [.090, .180] 0.075 0.924 0.924

Note. Quasi Equal Comm. = Quasi-Circumplex Model (Equal Communities Assumed); Quasi Equal Ang. = Quasi-

Circumplex (Equal Angular Location Assumed); Circumplex = Circumplex (or Circulant) Model; χ2 = chi-square;

df. = degree of freedom; RMSEA = root mean square error of approximation; SRMR = standardized root mean

square residual; CFI = comparative fit index; TLI = Tucker-Lewis NNFI.

* One parameter is on a boundary.

61

Table 14

Point Estimates and Confidence Intervals for Polar Angles and Communality

Group Estimate R I A S E C

Overall Polar Angle PE 0 349 260 225 159 108

CI [0, 0] [336, 2] [236, 284] [207,224] [130, 188] [89, 127]

Communality PE .84 .87 .62 .94 .76 .87

CI [.89, .89] [.81, .91] [.54, .69] [.83, .98] [.70, .82] [.76, .94]

Males Polar Angle PE 0 31 137 143 229 258

CI [0, 0] [5, 57] [92, 181] [115, 172] [191, 267] [226, 289]

Communality PE .84 .77 .60 1.00 .75 .81

CI [.69, .93] [.65, .87] [.47, .72] N/A [.64, .85] [.67, .90]

Females Polar Angle PE 0 11 333 261 213 137

CI [0, 0] [356, 26] [308, 357] [225, 297] [114, 283] [111, 163]

Communality PE .87 .89 .68 .90 .76 1.00

CI [.81, .91] [.83, .93] [.60, .76] [.77, .96] [.68, .83] N/A

Students Polar Angle PE 0 17 117 144 215 262

CI [0, 0] [0, 34] [79, 155] [120, 168] [179, 251] [238, 287]

Communality PE .79 .90 .51 1.00 .76 .87

CI [.69, .87] [.79, .96] [.39, .63] N/A [.66, .84] [.71, .95]

Employees Polar Angle PE 0 355 83 152 196 236

CI [0, 0] [328, 22] [38, 127] [100, 205] [128, 264] [198, 273]

Communality PE .84 .87 .73 .87 .73 .96

CI [.75, .91] [.78, .93] [.60, .83] [.75, .94] [.63, .82] [.61, 1.00]

Note. PE = point estimate; CI = 95% confidence interval.

62

Appendix B: Figures

Figure 1. Categorization of four specific Holland’s models

63

Figure 2. Visual presentation of four specific Holland’s models

64

Figure 3. Unconstrained (circular ordering) model for the overall sample (N = 364)

65

Figure 4. Unconstrained (circular ordering) model for the male group (N = 135)

66

Figure 5. Unconstrained (circular ordering) model for the female group (N = 229)

67

Figure 6. Unconstrained (circular ordering) model for the student group (N = 191)

68

Figure 7. Unconstrained (circular ordering) model for the employee group (N = 127)

69

Appendix C: IRB Review

Generalization of the Strong Interest Inventory in Chinese ...

Documents

Transcript of Generalization of the Strong Interest Inventory in Chinese ...